prospection test

library(tidyverse)
library(ggtext)
library(cowplot)

patients <- readxl::read_xlsx("2._byetta_insulin_music_diagram_v5 (30) (12).xlsx")
patients %>% 
  head()

patients <- patients %>% 
  pivot_longer(-`Patient ID`, names_to = "date", values_to = "med") %>% 
  rename(id = `Patient ID`)
patients

patients %>% 
  # filter(id == 1) %>%
  mutate(date = as.Date(as.numeric(date), 
                    origin = '1899-12-30')) %>% 
  group_by(id, date) %>% 
  arrange(id, date) %>% 
  mutate(bought_ins = ifelse("In" %in% med, 1, 0),
         bought_by = ifelse("By" %in% med, 1, 0)) %>% 
  filter(row_number() == 1) %>%
  select(-med) %>%
  ungroup() %>% 
  group_by(id) %>%
  mutate(used_ins = ifelse(bought_ins == 0 &
                             ((lag(bought_ins, default = 0, n = 1) == 1) |
                                (lag(bought_ins, default = 0, n = 2) == 1) |
                                (lag(bought_ins, default = 0, n = 3) == 1) |
                                (lag(bought_ins, default = 0, n = 4) == 1) |
                                (lag(bought_ins, default = 0, n = 5) == 1) |
                                (lag(bought_ins, default = 0, n = 6) == 1)),
                           1, bought_ins),
         used_by = ifelse(bought_by == 0 &
                             ((lag(bought_by, default = 0, n = 1) == 1) |
                                (lag(bought_by, default = 0, n = 2) == 1)),
                           1, bought_by),
         n_med = used_ins + used_by) %>% 
  mutate(comed = ifelse(used_ins == 1 & used_by == 1, 1, 0),
         comed_rate = round(sum(comed) / n() * 100,1)) %>% 
  summarise(comed_rate = round(sum(comed) / n() * 100,1)) %>%
  arrange(-comed_rate)

comed <- patients %>% 
  # filter(id == 1) %>%
  mutate(date = as.Date(as.numeric(date), 
                    origin = '1899-12-30')) %>% 
  group_by(id, date) %>% 
  arrange(id, date) %>% 
  mutate(bought_ins = ifelse("In" %in% med, 1, 0),
         bought_by = ifelse("By" %in% med, 1, 0)) %>% 
  filter(row_number() == 1) %>%
  select(-med) %>%
  ungroup() %>% 
  group_by(id) %>%
  mutate(used_ins = ifelse(bought_ins == 0 &
                             ((lag(bought_ins, default = 0, n = 1) == 1) |
                                (lag(bought_ins, default = 0, n = 2) == 1) |
                                (lag(bought_ins, default = 0, n = 3) == 1) |
                                (lag(bought_ins, default = 0, n = 4) == 1) |
                                (lag(bought_ins, default = 0, n = 5) == 1) |
                                (lag(bought_ins, default = 0, n = 6) == 1)),
                           1, bought_ins),
         used_by = ifelse(bought_by == 0 &
                             ((lag(bought_by, default = 0, n = 1) == 1) |
                                (lag(bought_by, default = 0, n = 2) == 1)),
                           1, bought_by),
         n_med = used_ins + used_by) %>% 
  mutate(comed = ifelse(used_ins == 1 & used_by == 1, 1, 0),
         comed_rate = round(sum(comed) / n() * 100,1),
         name = paste0("#",id, " (", comed_rate, "%)")) %>%  
  ungroup()
comed 

mycolors <- c("darkgoldenrod3", "coral3")
break.vec <- c(seq(from = as.Date("2012-01-01"), to = as.Date("2014-03-01"),
                 by = "6 months"))
ggplot(comed, aes(x=date, )) +
  geom_col(aes(y=comed, col = "Both"), fill = "gray40", alpha = 0.3, width = 14, show.legend = F) +
  geom_col(aes(y=-comed, col = "Both"), fill = "gray40", alpha = 0.3, width = 14, show.legend = F) +
  geom_vline(xintercept = as.numeric(as.Date(c("2012-01-01", "2013-01-01", "2014-01-01"))),
             col = "gray20", lwd = 0.3) +
  geom_line(aes(y = -used_ins, col="Insulin"), size = 1.2) +
  geom_line(aes(y = used_by, col = "Byetta"), size = 1.2) +
  geom_point(aes(y = -bought_ins * 2 + 1.8, fill = "Insulin"), col = mycolors[2], size = 0.8) +
  geom_point(aes(y = bought_by * 2 - 1.8, fill = "Byetta"), col = mycolors[1], size = 0.8, show.legend = F) +
  scale_fill_manual(name = "Bought:",
                    values = c(1, 1),
                    breaks = c("Byetta", "Insulin"),
                    guide = guide_legend(override.aes = list(color = mycolors,
                                                             size = 2))) +
  scale_color_manual(values = c(mycolors, alpha("white", 0)),
                     breaks = c("Byetta", "Insulin", "Both"),
                     guide = guide_legend(override.aes = list(color = c(mycolors, "gray80"),
                                                              size = c(2,2,6)))) +
  scale_y_continuous(breaks = c(-1,0,1), labels = c("Insulin", "None", "Byetta"), limits = c(-1,1)) +
  # scale_x_date(date_labels = "%b %Y", date_breaks = "6 months",
  #              minor_breaks = "1 month",
  #              expand = c(0,0),
  #              limits = c(min=min(date), max = max(date)),
  #              ) +
  scale_x_date(breaks = break.vec,
               date_labels = "%b %Y") +
  facet_wrap(id ~., labeller = as_labeller(setNames(comed$name, comed$id))) +
  labs(col = "Taking:", x = "Date", y = "Medication",
       title = "Insulin and Byetta Co-Medication Study", 
       subtitle = "Patient ID and co-medication rate (%), from January 2012 to February 2014") +
  theme_bw() +
  theme(axis.text.x = element_text(angle = 90, hjust = 1, vjust = 0.5),
        panel.grid.minor.y = element_blank(),
        strip.background = element_rect(fill = "gray98"))

comed %>% 
  mutate(id = fct_reorder(as.factor(id), comed_rate)) %>% 
  pull(id)

   [1] 1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1  1 
  [44] 1  1  1  1  1  1  1  1  1  1  1  1  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2 
  [87] 2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  2  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3 
 [130] 3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  3  4  4  4  4  4  4  4 
 [173] 4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4  4 
 [216] 4  4  4  4  4  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5 
 [259] 5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  5  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6 
 [302] 6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  6  7  7  7  7  7  7  7  7  7  7  7  7  7  7 
 [345] 7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  7  9  9 
 [388] 9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9  9 
 [431] 9  9  9  9  9  9  9  9  9  9  11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11
 [474] 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 11 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13
 [517] 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 14 14 14 14 14 14 14 14 14
 [560] 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14
 [603] 14 14 14 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15
 [646] 15 15 15 15 15 15 15 15 15 15 15 15 15 15 15 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16
 [689] 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 16 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17
 [732] 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 19 19 19 19
 [775] 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19
 [818] 19 19 19 19 19 19 19 19 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20
 [861] 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21
 [904] 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 22 22 22 22 22 22 22 22 22 22 22
 [947] 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22 22
 [990] 22 23 23 23 23 23 23 23 23 23 23
 [ reached getOption("max.print") -- omitted 2740 entries ]
68 Levels: 22 23 68 69 16 17 28 33 3 47 54 73 44 57 64 74 78 26 39 53 14 20 31 32 71 42 29 60 63 35 46 9 48 58 65 30 56 7 1 4 45 ... 41

mycolors <- c("darkgoldenrod3", "coral3")
break.vec <- c(seq(from = as.Date("2012-01-01"), to = as.Date("2014-03-01"),
                 by = "6 months"))

ggplot(comed %>% mutate(id = fct_reorder(as.factor(id), -comed_rate)), aes(x=date, )) +
  geom_col(aes(y=comed, col = "Both"), fill = "gray40", alpha = 0.3, width = 14, show.legend = F) +
  geom_col(aes(y=-comed, col = "Both"), fill = "gray40", alpha = 0.3, width = 14, show.legend = F) +
  geom_vline(xintercept = as.numeric(as.Date(c("2012-01-01", "2013-01-01", "2014-01-01"))),
             col = "gray20", lwd = 0.3) +
  geom_line(aes(y = -used_ins, col="Insulin"), size = 1.2) +
  geom_line(aes(y = used_by, col = "Byetta"), size = 1.2) +
  geom_point(aes(y = -bought_ins * 2 + 1.8, fill = "Insulin"), col = mycolors[2], size = 0.8) +
  geom_point(aes(y = bought_by * 2 - 1.8, fill = "Byetta"), col = mycolors[1], size = 0.8, show.legend = F) +
  scale_fill_manual(name = "Bought:",
                    values = c(1, 1),
                    breaks = c("Byetta", "Insulin"),
                    guide = guide_legend(override.aes = list(color = mycolors,
                                                             size = 2))) +
  scale_color_manual(values = c(mycolors, alpha("white", 0)),
                     breaks = c("Byetta", "Insulin", "Both"),
                     guide = guide_legend(override.aes = list(color = c(mycolors, "gray80"),
                                                              size = c(2,2,6)))) +
  scale_y_continuous(breaks = c(-1,0,1), labels = c("Insulin", "None", "Byetta"), limits = c(-1,1)) +
  scale_x_date(breaks = break.vec,
               date_labels = "%b %Y") +
  facet_wrap(id ~., labeller = as_labeller(setNames(comed$name, comed$id))) +
  labs(col = "Taking:", x = "Date", y = "Medication",
       title = "Insulin and Byetta Co-Medication Study", 
       subtitle = "Patient ID and co-medication rate (%), from January 2012 to February 2014") +
  theme_bw() +
  theme(axis.text.x = element_text(angle = 90, hjust = 1, vjust = 0.5),
        panel.grid.minor.y = element_blank(),
        strip.background = element_rect(fill = "gray98"))

mycolors <- c("darkgoldenrod3", "coral3")
mycolors <- c("#90EF77", "#017182") #9aee84, #073A52
highlight <- "coral3" #"#FC1B51"
break.vec <- c(seq(from = as.Date("2012-01-01"), to = as.Date("2014-03-01"),
                 by = "6 months"))

pdata <- comed %>% filter(id < 500) %>% mutate(id = fct_reorder(as.factor(id), -comed_rate))
ggplot(pdata, aes(x=date, )) +
  geom_col(data = pdata %>% filter(n_med <= 1),
           aes(y=n_med, col = "Both"), fill = "gray40", alpha = 0.5, width = 14, show.legend = F) +
  # geom_line(aes(y=n_med), col="black", show.legend = F) +
  geom_col(data = pdata %>% filter(n_med > 1),
           aes(y=n_med, col = "Both"), fill = highlight, alpha = 0.6, width = 14, show.legend = F) +
  geom_vline(xintercept = as.numeric(as.Date(c("2012-01-01", "2013-01-01", "2014-01-01"))),
             col = "gray20", lwd = 0.3) +
  # geom_line(aes(y = -used_ins, col="Insulin"), size = 1.2) +
  # geom_line(aes(y = used_by, col = "Byetta"), size = 1.2) +
  geom_point(data = pdata %>% filter(used_ins == 1, n_med == 1),
             aes(y = 0.1, fill = "Insulin"), col = mycolors[1], size = 0.6) +
  geom_point(data = pdata %>% filter(used_by == 1, n_med == 1),
             aes(y = 0.1, fill = "Byetta"), col = mycolors[2], size = 0.6) +
  geom_point(aes(y = -1, fill = "Both"),) +
  # geom_point(aes(y = bought_by * 2 - 1.8, fill = "Byetta"), col = mycolors[1], size = 0.8, show.legend = F) +
  scale_fill_manual(name = "Medication:",
                    values = c(1, 1, 1),
                    breaks = c("Insulin", "Byetta", "Both"),
                    guide = guide_legend(override.aes = list(color = c(mycolors, highlight),
                                                             shape = c(16,16,15),
                                                             alpha = c(1,1,0.4),
                                                             size = c(2,2,6)))) +
  scale_color_manual(values = c(mycolors, alpha("white", 0)),
                     breaks = c("Byetta", "Insulin", "Both"),
                     guide = guide_legend(override.aes = list(color = c(mycolors, "gray80"),
                                                              size = c(2,2,6)))) +
  scale_y_continuous(breaks = c(0,1,2),
                     labels = c("0", "1", "2"),
                     limits = c(0, 2)) +
  scale_x_date(breaks = break.vec,
               date_labels = "%b %Y") +
  facet_wrap(id ~., labeller = as_labeller(setNames(comed$name, comed$id))) +
  labs(col = "Taking:", x = "Date", y = "# Taken Medications",
       title = "Insulin and Byetta Co-Medication Study", 
       subtitle = "Patient ID and co-medication rate (%), from January 2012 to February 2014") +
  theme_bw() +
  theme(axis.text.x = element_text(angle = 90, hjust = 1, vjust = 0.5),
        axis.text.y = element_blank(),
        axis.ticks.y = element_blank(),
        panel.grid.minor.y = element_blank(),
        strip.background = element_rect(fill = "gray98"))

pdata <- comed %>% filter(id < 500) %>% mutate(id = fct_reorder(as.factor(id), comed_rate)) %>% 
  mutate(med = case_when(
    n_med == 2 ~ "Both", 
    n_med == 1 & used_ins == 1 ~ "Insulin",
    n_med == 1 & used_by == 1 ~ "Byetta",
    TRUE ~ "None"
  ),
  alpha = ifelse(med == "None", 0, 1),
  name = paste0("#",id, " **(", comed_rate, "%)**"))
axis_size <- 5.5
p1 <- ggplot(pdata %>% filter(as.numeric(id) > 34), aes(x=date, y=id)) +
  geom_tile(aes(fill = factor(med), alpha = alpha), size = 1) +
  # geom_point(aes(col = factor(med), alpha = alpha), shape = 15, size = 5) +
  # geom_point(data = pdata %>% filter(used_ins == 1, n_med == 1),
  #            aes(fill = "Insulin"), col = mycolors[1], size = 0.6) +
  # geom_point(data = pdata %>% filter(used_by == 1, n_med == 1),
  #            aes(fill = "Byetta"), col = mycolors[2], size = 0.6) +
  scale_y_discrete(breaks = pdata$id,
                   labels = pdata$name) +
  scale_fill_manual(values = c(highlight, mycolors, "white"),
                     breaks = c("Both", "Insulin", "Byetta", "None"),
                     guide = guide_legend(override.aes = list(color = c("white", "gray80", highlight),
                                                              size = c(2,2,6)))) +
  scale_x_date(breaks = break.vec,
               date_labels = "%b %Y") +
  theme_bw() +
  labs(col = "Taking:", x = "Date", y = "Patient ID",
  ) +
  theme(legend.position = "none",
        axis.text.y = element_markdown(size=axis_size),
        plot.title = element_markdown(lineheight = 1.1),
        legend.text = element_markdown(size = 11)
        )
p2 <- ggplot(pdata %>% filter(as.numeric(id) <= 34), aes(x=date, y=id)) +
  geom_tile(aes(fill = factor(med), alpha = alpha), size = 1) +
  scale_y_discrete(breaks = pdata$id,
                   labels = pdata$name) +
  scale_fill_manual(values = c(highlight, mycolors, "white"),
                     breaks = c("Both", "Insulin", "Byetta", "None"),
                     guide = guide_legend(override.aes = list(color = c("white", "gray80", highlight),
                                                              size = c(2,2,6)))) +
  scale_x_date(breaks = break.vec,
               date_labels = "%b %Y") +
  theme_bw() +
  labs(col = "Taking:", x = "Date", y = "",
       title = "",
  ) +
  theme(legend.position = "none",
        axis.text.y = element_markdown(size = axis_size),
        plot.title = element_markdown(lineheight = 1.1),
        legend.text = element_markdown(size = 11)
        )
title <- ggdraw() + geom_richtext(
    data = data.frame(x = 0.03, y = 0.5, 
                      label = "<span style='font-size:14pt'>**Insulin and Byetta Co-Medication Study**</span><br>
    <span style='font-size:11pt'>Patient ID and **co-medication rate (%)**, from January 2012 to February 2014 <br> 
    Comparison between the use of <span style='color:#6BB058;'>**Insulin**</span>, 
    <span style='color:#017182;'>**Byetta**</span> or
    <span style='color:#cd5b45;'>**both**</span> medications at the same time
    </span>"),
    aes(x, y, label = label),
    fill = NA, label.color = NA,
    hjust = 0, vjust = 0.5, angle = 0,
    label.padding = grid::unit(rep(0, 4), "pt"),
    # color = "black",
    inherit.aes = FALSE
  )
plot_grid(title, plot_grid(p1,p2), nrow =2, rel_heights = c(0.2, 1))